Answer

Categorical Data Analysis: Chi-Square Test of Association

Categorical data are recorded counts of observations that fall into any particular bin or hole, rather than quality of the observations themselves. Categorical data are either measured at the Ordinal level (categories with some inherent order) or nominal scale (categories with no particular order). Therefore, categorical data analysis focuses on the bins rather than the observations themselves, an examination of the collection of observations. Therefore, among the assumptions anchoring this mode of analysis include levels of measure (either nominal or ordinal), data consists of two or more independent categories, and categories are partitions (the pigeons/holes in which observations are placed are mutually exclusive and exhaustive) (Frankfort-Nachmias & Leon-Guerrero, 2017). The criterion for handling categorical data in analysis also dictate that observations are not measured but rather their count of existence to certain holes/partitions/bins. This study is a categorical data analysis of certain categorical variables in the GSS dataset.

General Social Survey Dataset

There are various categorical variables in the dataset that can be used to assess certain attributes of the population. In this case, the variables “Political Party Affiliation” and “Sex of Respondent” are considered. The research question to be answered in this expedition is: Does gender determine in any level one’s political affiliation in America?

The null hypothesis to be tested states that there is no association between gender and political affiliation among Americans.

The descriptive study design fits this research question since the interest is in observation and description of characteristics or behaviours of the population, without examining the how/when/why questions about the behaviours, as articulated by Shuttleworth, (2008). The fit of the design is further grounded on the fact that causal effects of the association between gender and political affiliation are not required.

In this analysis “Respondent’s Sex” is the independent variable while “Political Party Affiliation” is the dependent variable. Both variables are measured on nominal levels, hence fitting categorical data analysis techniques in examination. 

 Bivariate Table

Table 1: Bivariate table of Race and {Political Party Affiliation (Rows and Columns switched to fit in the paper).

While the survey participation constituted more females than males (55% to 45%), females are preliminary results from the cross-tabulation shows that women are more politically dynamic than men. This insinuation is affirmed by the higher distribution of females in both Democratic and Republican Categories (Either strong or not strong), pointing out the possibility of females being more politically dynamic than their male counterparts. Males, however, lead in Independent parties inclined towards Republican and affiliation to other parties outside the two main parties.

The Pearson Chi-Square Test of Association

Table 2: Table of Chi-Square test of association (Special focus on the Pearson Chi-Square column)

From the dataset, the χ² = 24.95, p = .001 < 0.05, indicates that there is significant statistical association between gender and political affiliation (reject the null hypothesis). That is, males and females vary in political orientation.

 Table 3: test of association strength.

Phi and Cramer’s V are both tests of the strength of association. They both indicate significance (p < 0.05), meaning the association between gender and political affiliation is strong.

Discussion

While the little depth of previous research points to significant association between political affiliation and gender (a fact further affirmed by this study), they have mostly indicated that most women are more predisposed Democrat-oriented while men are more affiliated to the Republican Party (Rule & Ambady, 2010). In this replication, a difference is found – that women are more politically varied than men, in the sense that more women than men are in the two main political parties as further illustrated in the bar-graph below. Therefore, we conclude that based on the data, the female gender is more politically dynamic than its male counterpart.